1 Mock CMO

1. (Lithuania team contest 2000, #3)

Solve the equation
p p p
−3x2 + 18x + 37 + −5x2 + 30x − 41 = x2 − 6x + 109.

Solution: √ √ √
After setting t = (x − 3)2 , the equation reduces to 64 − 3t + 4 − 5t = t + 100. t = 0 is
one solution, and the left-hand side is decreasing in t while the right-hand side is increasing in t.
Therefore, t = 0 (and hence x = 3) is the unique solution.
2. (Ukraine 10th grade 2008, #2)
Let ABCD be a parallelogram with 6 BCD > 90◦ . Extend line BC to K so that DK = DC, and
extend line DC to L so that BL = BC. Let the bisectors of 6 CDK and 6 CBL meet at point Q. If
6 BQD = α and KL = a, express the circumradius of triangle AKL in terms of a and α.

Solution:
ADLB and ADKB are isosceles trapezoids, so K, L lie on the circumcircle of 4ABD. 6 BAD =
6 LCK = 180◦ − 6 BQD since BQ and QD are perpendicular to LC and CK. Therefore, ABLQKD
is a cyclic hexagon. Now, 6 LAK = 6 LAQ + 6 KAQ = 6 LDQ + 6 KBQ = 180◦ − 2α. Applying
the extended sine law, we have the radius is 2 sina 2α .
3. (Japan junior math olympiad 2008, #10)
Consider a board consisting of 9 squares of length 1, arranged in 3 rows and 3 columns as in the figure
below. Let us number squares as indicated in the figure.

1 2 3
8 4
7 6 5

Let A, B, C, D be points chosen from the interior (not on the boundary) of the square 1, 3, 5, 7,
respectively. Denote by X the sum of the areas of the intersection of the quadrilateral ABCD with the
squares 1, 3, 5, and 7. Denote also by Y the sum of the areas of the intersections of the quadrilateral
ABCD with the squares 2, 4, 6, and 8. Prove that X < Y .
Solution:
Suppose that ABCD intersects the boundary of square i and square i + 1 at point Xi , and that
Xi is distance xi from P8 the central square. Then, the area of ABCD within square 2i is exactly
x2i−1 +x2i 1
2 , so Y = 2 i=1 xi . Now let Yi and Zi denote the outer and inner corners of square i for
i = 1, 3, 5, 7. Then, the area of ABCD within square 2i − 1 is less than the area of quadrilateral
X2i−2 Y2i−1 X2i−1 Z2i−1 , which is exactly 1 − 1−x22i−2 − 1−x22i−1 = x2i−2 +x
2
2i−1
. Therefore, X <
1
P8
2 i=1 ix = Y .

4. (Lithuania team contest 2000, #10)

A positive integer n is said to be reducible if there exist positive integers m and d such that
m+1 m+2
n= + .
d+1 d+2
How many reducible numbers are there from the set {1, 2, 3, . . . , 2000}?
Solution:
m+1 m+2 (m+1)(d+2)+(m+2)(d+1)
d+1 + d+2 = (d+1)(d+2) . For this to be an integer, we must have (m + 2)(d + 1) ≡
0 (mod d + 2) =⇒ m + 2 ≡ 0 (mod d + 2) and (m + 1)(d + 2) ≡ 0 (mod d + 1) =⇒ m + 1 ≡

1
 0 (mod d + 1). Equivalently m ≡ d (mod lcm(d + 1, d + 2)). Since d + 1 and d + 2 are relatively
prime, it follows that m = C(d + 1)(d + 2) + d, which gives n = C(2d + 3) + 2.
Clearly, n = 2 can be expressed in this form but n = 1 cannot. For n > 2, the condition is
equivalent to requiring that n − 2 have an odd divisor greater than 3. This fails only for n = 2 + 2k
and n = 2 + 3 · 2k . Altogether, that gives only 1 + 11 + 10 numbers that are not reducible, so the
answer is 1978.
5. (Japan math olympiad final 2008, #5)
Does there exist a positive integer n satisfying the following condition?
Condition: For an arbitrary rational number r, there exists an integer b and nonzero integers a1 , a2 , . . . , an
such that r = b + a11 + . . . + a1n .
Solution:
No. We prove by induction on n that for each integer n and rational number q, there exists a
rational number q 0 < q such that no number in the interval (q 0 , q) can be expressed in this form.
For n = 1, the claim is trivial.
Now suppose the claim has been proven for n − 1. Fix an arbitrary q, and let f denote the
fractional part of q (except if q is an integer, we take f = 1). Then, if we are to express some
number y ∈ (q − f2 , q) in the required form, we must have some a1i ≥ f2 =⇒ ai ≤ f2 . This leaves
only a finite number of options for ai . Once ai is fixed to some value u, we have reduced the problem
to the n − 1 case, so by the inductive hypothesis, there exists qu0 such that no number in (qu0 , q) can
be achieved, given that ai = u. The inductive step follows from taking q 0 = max(q − f2 , maxu qu0 ).

2
2 Mock IMO Day 1
1. (Ukraine 11th grade 2008, #3)
A point O is placed inside triangle ABC so that 6 BOC = 90◦ and 6 BAO = 6 BCO. If M and N are
the midpoints of the segments AC and BC respectively, prove that 6 OM N = 90◦ .
Solution:
Let P be the midpoint of OC. N is the circumcenter of right triangle BOC since it is the midpoint
of the hypoteneuse, so 6 BCO = 6 N CP = 6 N OP . Since M P is parallel to AO and M N is parallel
to AB, we have 6 N M P = 6 BAO = 6 BCO = 6 N OP . Therefore, OM P N is a cyclic quadrilateral
and 6 OM N = 6 OP N = 6 BOC = 90◦ .
2. (Japan math olympiad final 2008, #4)
Determine all real-valued functions f defined on the real line, which satisfy

f (x + y)f (f (x) − y) = xf (x) − yf (y).

for all real numbers x and y.

Solution:
Taking y = 0 gives f (x)f (f (x)) = xf (x), so f (x) = 0 or f (f (x)) = x. Taking y = f (x) gives
f (x + f (x))f (0) = xf (x) − f (x)f (f (x)) = 0. Taking y = −x gives 0 = f (0)f (f (x) + x) =
xf (x) + xf (−x), which implies f (−x) = −f (x) for x 6= 0.
Now suppose f (x) = 0 and y, f (y) 6= 0. Then, f (x+y)f (−y) = −yf (y) = yf (−y) =⇒ f (x+y) = y.
In particular, f (x + y) 6= 0 so f (f (x + y)) = x + y, and f (y) = x + y (*). If x 6= 0, then f (−x) = 0
also and the same argument implies f (y) = −x + y, which is impossible. Therefore, if f (x) = 0 for
some x 6= 0, then f (x) = 0 for all x 6= 0. It is easy to check any such function is a solution.
Otherwise, f (x) 6= 0 for all x = 0. Then f (0)f (f (0)) = 0 implies f (0) = 0, and (*) implies that in
this case, f (y) = y for all y. This, in turn, is easily checked to be a valid solution.
3. (Bulgarian math olympiad team selection test 2008, #1)
The number -1 is written at k of the vertices of a regular 2009-gon and the number 1 is written at
the remaining 2009 − k vertices. A vertex is said to be good if, starting from this vertex and running
around the polygon in either direction, every partial sum is positive. Find the largest number k such
that there exists a good vertex for any arrangement of the 1’s and -1’s.
Solution:
Let p be the number of positive vertices, and n the number of negative vertices. If p ≤ 2n, put all
positive vertices together. Then, any positive vertex can reach the n negative vertices after going
through at most n positive vertices, so it is bad. Conversely, if p > 2n, then for each positive vertex,
there must exist a minimal interval with it as an endpoint and sum 0. The opposite endpoint is
negative, and each negative point is endpoints to two such intervals. Therefore, some vertex is
good. k = 669.

3
3 Mock IMO Day 2
4. (IMO short list 2008, N1) Let n be a positive integer and let p be a prime number. Prove that if a, b, c
are integers (not necessarily positive) satisfying the following equations

an + pb = bn + pc = cn + pa,

then a = b = c.
Solution:
If two values are equal, the claim is trivial. Otherwise, multiply the equations an − bn = p(c − b) to
n
−bn bn −cn cn −an
get a a−b · b−c · c−a = −p3 . If n is odd, an − bn has the same sign as a − b, and the left-hand
side is positive, which is impossible. Otherwise, two of a, b, c are the same parity, and in this case,
n
−bn
the corresponding a a−b term is even, which implies p = 2.
n n
−b
The original equation now implies a, b, c must all be the same parity, and hence a a−b is even (as
3
above). Since the product of these terms is −2 , each such term is ±2. Comparing with the original
equation, we have a − b = ±(b − c). If a − b, b − c, c − a are all the same sign, they are all equal,
but their sum is 0 so then a = b = c. Otherwise, a − b = c − b or some shift thereof, and two values
are equal. Either way, we have a contradiction.
5. (Ukraine 11th grade 2008, #7)
Prove that the inequality
x y z √
p +p +p ≤ 3
x2 + y + z x + y2 + z x + y + z2

holds for any non-negative real numbers x, y, z satisfying x2 + y 2 + z 2 = 3.

Solution:
x(1+y+z)+y(1+z+x)+z(1+x+y)
The given condition implies x+y +z ≤ x2 +y 2 +z 2 = 3, and hence (x+y+z)2 ≤ 1.
1 1+y+z x y z
By Cauchy-Schwarz, x2 +y+z ≤ (x+y+z) 2 , so therefore, x2 +y+z + x+y 2 +z + x+y+z 2 ≤ 1. The result
now follows from x + y + z ≤ 3 and Cauchy-Schwarz.
6. (Romanian master in mathematics 2008, #4)
Prove that from among any (n + 1)2 points inside a square of side length positive integer n, one can
pick three determining a triangle with area at most 12 .
Solution:
If three of the N = (n + 1)2 points are collinear, the problem is trivial. Otherwise, let H denote the
convex hull of the points, and let k denote the number of points on the boundary of H. Then, H has
area at most n2 and perimeter at most 4n. We first triangulate H. This uses exactly 2(N − 1) − k
n2
triangles, so one triangle has area at most 2(N −1)−k . Next, there must exist consecutive sides of
H with lengths a, b such that a+b 2 ≤ 4n
k . The area of the triangle with these two sides is at most
1 1 a+b 2 8n2


2 · ab ≤ 2 · 2 ≤ k2 .
 
n2 2
Therefore, some triangle has area at most X = min 2(N −1)−k , 8n
k2 . The first term is increasing in
n2 8n2
k, and the second is decreasing, so X is maximized when 2(N −1)−k = k2 , or equivalently k = 4n.
In this case, X = 21 , and the result follows.